Radio Broadcast (May 1928-Apr 1929)

Z RADIO BROADCAST the tTEinsformer leakage reactance. Since the transformers employed high-permeability cores (about five times the permeability of silicon steel cores) the leakage reactance was low, and the resonant frequency was high — well above the usual range of frequencies transmitted by broadcast stations. This is a very good characteristic. After completing this test there were a number of experiments which we had to perform. "How much filtering can we remove before the characteristic goes bad?" "Or will it go bad?" These were questions which we had to answer. First, we removed Ri, and connected the 22.5 volts directly from the B supply to the lower end of the primary of Ti This left 1.0 mfd. across the 22.5-volt tap. At 60 cycles (C) of Fig. 5 the amplifier went down 6.6 db from its 1000-cycle level, and the entire amplifier was down 1.0 db. Why? Those who read the "Armchair Engineer" (March, 1928, Radio Bboadcast) will remember we calculated the amount of a.c. current that flowed through a 40-henry choke (which is a good big choke) when the loud speaker was connected as in this test. This a.c. current, although small, flows through the B supply, sets up an a.c. voltage there, and part of this voltage appears across the 22.5-volt tap. This voltage is fed into the input of the amplifier and, of course, is amplified. Since the amplifier was now down at 60 cycles, it means that the transformers were so "poled" that this a.c. voltage impressed on the input from the B supply, returned to the 4000-ohm output resistor out of phase with the original voltage, and hence was subtracted from it. It is worth while, then, to filter the detector. In this case, despite the 1.0-mfd. condenser across the 22.5-volt tap, the amplifier was down, indicating that this capacity (which has a reactance of 2650 ohms at 60 cycles) was too small to give any bypassing effect. When the resistor, Ri, of 50,000 ohms, was in the circuit, however, the condenser was relatively much more effective and kept these a.c. voltages from entering the primary of the first a.f. transformer. I Other Experiments N THE next experiment we increased the current through the B supply by connecting the output resistor directly across output choke, thereby reducing its impedance to approximately 4000 ohms and increasing the a.c. through the B supply by the amount that previously went directly to the filament. Now Fig. 4 — Top view of completed amplifier shows exact arrangement of apparatus on baseboard the amplifier (D) in Fig. 5 was down 14.8 db at 60 cycles, and down 7.0 db at 1000 cycles. Placing 8 mfd. across the Majestic unit increased the 60-cycle response to 12.8 db — a gain of 2 db — but this was not worth while. Since this is analogous to operating the amplifier with an output transformer, it is absolutely essential that the detector supply be well filtered and as near (physically) the first a.f. transformer as possible. This necessitates two wires from the detector to the amplifier as in Fig. 1. Filtering the 22.5-volt circuit (the detector platevoltage supply), as in Fig. 1, and removing all filtering from the first a.f. tube, provided a good characteristic, almost as good as with the filtering in the circuit. Taking out all the filtering, even the by -pass condensers across the C-bias resistors, gives the characteristic shown at (B) in the curve. With this arrangement the amplifier suffered badly at both low and high frequencies. Placing the loud speaker across the choke and removing all filtering gave the characteristic at (E) — which in our estimation is pretty terrible. Replacing the condenser across the C bias to the last tube increased the 1000-cycle response to normal but improved the 60 and 100-cycfe response only 4 db, showing that this capacity is far too small to do much good compared 10 000 ohms 12.000 ohms -B Ground 6+B Fig. 3 — Schematic diagram of a well-filtered two-stage transformercoupled a.f. amplifier with the 2000-ohm resistor which it bypasses. In other words, at low frequencies its reactance is too great to be of much good as a bypassing agent. It is only when the circuit beyond this condenser (toward the B supply) is increased in impedance (one-half megohm, Rs) that the condenser gets in its good licks. Now with all the filtering in place, a 1000ohm resistor placed in the B-supply lead caused no change in the characteristic at either low or high frequencies, proving that if the amplifier is properly constructed, it is independent of the source of plate or grid voltages. Conclusive Proof HERE, then, is conclusive proof that Mr. Crom is correct. The audio amplifier must be filtered if the good characteristic of a single transformer is to be preserved when a twostage affair is constructed. Here is an amplifier operating entirely from a.c, that gives a flat frequency characteristic from 100 to 8000 cycles and a power output of roughly 100 milliwatts for an input r.m.s. signal of 0.1 volt, or to put it another way, an amplifier that requires an input r.m.s. signal of 0.31 volts to produce a one-watt output. This is true providing the 171-type tube with a mu of 3 is used and providing it works into twice its own internal plate resistance. It is a beautiful amplifier, since it has not only a good characteristic but plenty of overall gain as well. Now how much of this filtering is necessary? So far as this particular amplifier and particular B supply are concerned, we can do away with the filtering in the grid and plate circuits of the first audio tube — but it is highly questionable whether such an economy would be a true saving. The filtering in the detector plate lead and the shunt condenser and series resistance in the grid circuit of the power tube are both absolutely essential. But so far as a general amplifier and a general B supply are concerned, we need every bit of filtering there is in this present assembly of apparatus as shown in Fig. 3. While it is true that the ear will not detect differences of perhaps 10 db at the two extremes of the audio band, it is not safe to say that we can eliminate such filtering as is only effective in bringing up these frequencies. If, by accident, one of the transformer primaries february, 1929 page 251 #